I would be more inclined to DARTEL all three datasets together, than do them seperately and try to join the data together afterwards. Also, make sure you use the latest updates as there was a fix made to the template generation part. Best regards, -John ---------------------------------------------------------------------------- Dear John, we have three large (each >200) high resolution T1 datasets that should ideally be joined for a large VBM analysis. The differ slightly in voxel size and contras and low frequency intensity bias; each of them is (for itself) of sufficient quality for VBM. For now, I DARTELed all three datasets separately, planning to introduce a covariate for the three different datasets later to correct for systemic differences between the three raw data types. I have two questions: 1. Would you (after all three datasets were brought to MNI space by affine normalisation of the respective 6th gen. template with MNI template) still do some adjustment/coregistration between the three templates? 2. Would joining all datasets irrespective of their different raw data features in one DARTEL process be okay? I guess systemic differences of the resulting modulated images could still be there but corrected in the model then. It is more to optimize cortical alignment. However, I am not sure if e. g. the different bias are better corrected if the datasets are handled separately first. Thanks a lot for any comments/support! best regards, Philipp